TorqueField is a physics-reasoning foundation model for machines operating across critical industries, from agriculture and heavy equipment to airports and infrastructure. Built to perform in dynamic, unpredictable, and harsh outdoor conditions.
TorqueField deploys across tractors, harvesters, sprayers, excavators, drills, boom systems, construction equipment, mining machines, drones, mobile platforms, and fixed site infrastructure. It reasons across RGB cameras, thermal imagers, LiDAR, radar, and multispectral sensors, and runs directly at the edge on NVIDIA Orin, NVIDIA Thor, RTX GPUs, and custom edge SoCs.
TorqueField combines perception, physical understanding, and spatio-temporal reasoning into a world model of the site it operates on, tracking how terrain, weather, and materials evolve while work happens. That model is what carries across new crops, sites, and operations with a fraction of the task-specific data.
Run TorqueField as a complete stack, or integrate single components: perception, physical reasoning, or planner cost functions. Each one fits an existing architecture on its own, without retraining the full model.
A world model of the site that holds terrain, contact, and material behavior together. It reasons through the interaction before the machine commits to an action.
Real-time inference on the equipment itself, so operation stays independent of connectivity across large and remote sites. Available as ready-to-use hardware kits.
Tell us what you're running and the workflow you want to crack — we'll deliver a model tuned to your deployment and our team will reach out within one business day.
Get the latest on how generative AI powers real-world robotics—no noise—just insights, updates, and breakthroughs.
